Skip to content

HLT and L1 Trigger Rate Monitoring Tools

Notifications You must be signed in to change notification settings

cms-tsg-fog/RateMon

Repository files navigation

RateMon

This repository provides various tools to monitor HLT and L1 rates. More details on the training repository.

GitHub & GitLab This repository is both on Github and CERN GitLab, to allow the usage of the gitlab CI/CD tools. Do any development work on GitLab.

Development

Requirements

To run RateMon, you need Python 3 and a couple of system packages required by the OMSAPI client. On Alma/RHEL systems, those can be satisfied installing the krb5-devel and python3-devel packages.

yum install krb5-devel python3-devel

To set up a Python virtual environment, activate it and install the python dependencies, run the following commands:

python3 -m venv venv
source venv/bin/activate
python3 -m pip install -r requirements.txt

CI/CD

Each push to GitLab that introduces new commits will start a CI pipeline. The source for that is the .gitlab-ci.yml file.

Each pipeline contains a build of an RPM package for the RateMon tools. These packages are automatically uploaded to a shared EOS folder.

Testing and Deploying Changes to RateMon

Tests to complete before creating a RateMon MR

Before creating a MR, thoroughly test the new code with the following tests. All of these tests should be done after a git pull and a git checkout of the new branch has been done on the machine so that the test runs on the new code.

Miscellaneous checks

  1. Check that in ShiftMonitorNCR.py, self.sendMattermostAlerts_static is set to True. If this is false, then alerts will not be sent to the RateMon alerts channel on Mattermost.

Tests on lxplus

  1. Run plotTriggerRates.py: python3 plotTriggerRates.py --triggerList=TriggerLists/monitorlist_COLLISIONS.list 370725
  2. Run ShiftMonitorTool.py without the config file: python3 ShiftMonitorTool.py --simulate=370725
  3. Run ShiftMonitorTool.py with the config file: python3 ShiftMonitorTool.py --simulate=370725 --configFile=ShiftMonitor_config.json

Tests on VM (caer)

  1. Run plotTriggerRates.py: python3 plotTriggerRates.py --triggerList=TriggerLists/monitorlist_COLLISIONS.list 370725
  2. Run test queries from the web interface (http://caer.cern.ch/api/v1/ui), with all different combintations of queryByFill and createFit, while setting the runNumber to the corresponding run/fill number for a given triggerKey. Preferably, use a recent run/fill and a standard trigger. Make sure the expected output (JSON, ROOT) is produced while monitoring the output on the VM.

Tests on P5 dev VM (kvm-s3562-1-ip149-08)

NOTE: These instructions need to be updated as per Issue #120

Login to machine:

ssh lxplus
ssh cmsusr
ssh kvm-s3562-1-ip149-08

This should put you into the directory /nsfhome0/USERNAME. If it is your first time using the machine, setup the ratemon repository here using git clone https://gitlab.cern.ch/cms-tsg-fog/ratemon.git.

Navigate to the ratemon directory and do a git pull and git checkout for the branch you want to test. Then run the following two tests:

  1. Test make_plots_for_cron_manual.py using a test fill (This is not available yet. In the meantime manually edit make_plots_for_cron.py to run over a specified fill by commenting out the line run_lst , fill_num = parser.getRecentRuns() and replacing it with two lines: fill_num = 9068 and run_lst = parser.getFillRuns(fill_num)).
  2. Test ShiftMonitorTool on the dev machine via a systemctl process:
sudo systemctl start ratemon.service
sudo journalctl -fu ratemon
sudo systemctl stop ratemon.service

Deployment

GitLab CI can deploy to P5. To do this, perform the following steps:

  • Create a new tag
  • Create a new pipeline
    • Run for must be set to your newly created tag instead of master
    • Set variables P5_USER and P5_PASS to your P5 username and password (note that P5 username is the same as CERN username, but P5 password is cmsusr password).
  • In this newly created pipeline, press the deploy:P5 play button
    • In order to manually deploy successfully, you also need to be on the cms ratemon librarian group.

After Deploying Changes

After deploying changes to the P5 machine, do the following to update the ater machine and confirm everything is running as expected.

  1. Check that the P5 machine cron job is running correctly. First, login to the P5 prod machine (kvm-s3562-1-ip151-84) and then view the cron job logs with the command sudo vim /var/spool/mail/hltpro (with your text editor of choice). The cron job only runs once an hour so this needs to be checked once the cron job for the new RateMon tag is running.
  2. Check the ShiftMonitorTool output logs. On the P5 prod machine, run this command to monitor the logs and confirm there are no errors: sudo journalctl -fu ratemon.
  3. Update the ater VM. Login to the ater machine and connect to the tmux session with tmux a -t 0. Then do CTRL+C to quit the current server. Next, do a git pull and checkout the new tag. Restart the API with python3 server.py. Then open https://cmsoms.cern.ch/ratemon/rate_vs_pu/ and click on a rate vs pu plot. If the API is working correctly, there will then be a corresponding interactive "live" plot.

Running ShiftMonitorTool

At P5, ratemon is installed for you and available as the ratemon Systemd service. This is on the VMs kvm-s3562-1-ip151-84 (production) and kvm-s3562-1-ip149-08 (development), which can be accessed from cmsusr.

To view logs:

journalctl -fu ratemon

To restart the Systemd service (done if there is an error or after new deployment) it can be done with:

sudo systemctl stop ratemon.service
sudo systemctl start ratemon.service

To run outside P5:

cd ratemon
source venv/bin/activate
python3 ShiftMonitorTool.py

About

HLT and L1 Trigger Rate Monitoring Tools

Resources

Stars

Watchers

Forks

Languages